Surrogate-based aerodynamic optimization of proprotor blades for a parcel delivery UAV

This study presents a surrogate-based aerodynamic optimization framework for a quad-tiltrotor platform designed for both hover and forward flight. To obtain an optimized blade design in both hover and forward flight conditions, a free wake model is coupled with Gaussian Process Regression (GPR). The framework enables efficient exploration of the design space while maintaining physical fidelity in rotor modeling. The rotor blade geometry is parameterized using piecewise linear functions for twist and taper distributions, allowing localized shape control during optimization. Latin Hypercube Sampling (LHS) is used to generate initial design points, and the surrogate model is iteratively refined using the Expected Improvement criterion. The final optimized blade exhibits a 3.7% improvement in hover figure of merit and a 38.3% increase in cruise propulsive efficiency compared to the baseline blade. These gains result from redistributed twist and chord distributions that enhance spanwise loading and reduce induced power loss. The results demonstrate the efficacy of combining free wake analysis with surrogate modeling for the design of UAV-scale rotorcraft.